microorganisms Article Variation in Human Milk Composition Is Related to Differences in Milk and Infant Fecal Microbial Communities Ryan M. Pace 1,* , Janet E. Williams 2, Bianca Robertson 3,4, Kimberly A. Lackey 1, Courtney L. Meehan 5, William J. Price 6 , James A. Foster 7, Daniel W. Sellen 8 , Elizabeth W. Kamau-Mbuthia 9, Egidioh W. Kamundia 9, Samwel Mbugua 9, Sophie E. Moore 10,11, Andrew M. Prentice 11 , Debela G. Kita 12, Linda J. Kvist 13, Gloria E. Otoo 14, Lorena Ruiz 15,16, Juan M. Rodríguez 17 , Rossina G. Pareja 18, Mark A. McGuire 2, Lars Bode 3,4 and Michelle K. McGuire 1,* 1 Margaret Ritchie School of Family and Consumer Sciences, University of Idaho, Moscow, ID 83844, USA; kimberlyl@uidaho.edu 2 Department of Animal, Veterinary and Food Sciences, University of Idaho, Moscow, ID 83844, USA; janetw@uidaho.edu (J.E.W.); mmcguire@uidaho.edu (M.A.M.) 3 Larsson-Rosenquist Foundation Mother-Milk-Infant Center of Research Excellence, Univeristy of California San Diego, La Jolla, CA 92093, USA; bmarieinsd@gmail.com (B.R.); lbode@health.ucsd.edu (L.B.) 4 Department of Pediatrics, Univeristy of California San Diego, La Jolla, CA 92093, USA 5 Department of Anthropology, Washington State University, Pullman, WA 99164, USA; cmeehan@wsu.edu 6 Statistical Programs, College of Agricultural and Life Sciences, University of Idaho, Moscow, ID 83844, USA; bprice@uidaho.edu  7 Department of Biological Sciences, University of Idaho, Moscow, ID 83844, USA; jamesafoster@mac.com  8 Department of Anthropology, University of Toronto, Toronto, ON M5S 1A8, Canada; dan.sellen@utoronto.ca Citation: Pace, R.M.; Williams, J.E.; 9 Department of Human Nutrition, Egerton University, Nakuru 20115, Kenya; Robertson, B.; Lackey, K.A.; Meehan, ekambu@yahoo.com (E.W.K.-M.); egidioh.kamundia@egerton.ac.ke (E.W.K.); C.L.; Price, W.J.; Foster, J.A.; Sellen, samwel.mbugua2@gmail.com (S.M.) 10 D.W.; Kamau-Mbuthia, E.W.; Department of Women and Children’s Health, King’s College London, London WC2R 2LS, UK; sophie.moore@kcl.ac.uk Kamundia, E.W.; et al. Variation in 11 MRC Unit The Gambia at the London School of Hygiene and Tropical Medicine, Fajara P.O. Box 273, Gambia; Human Milk Composition Is Related Andrew.Prentice@lshtm.ac.uk to Differences in Milk and Infant 12 Department of Anthropology, Hawassa University, Hawassa P.O. Box 27601, Ethiopia; debelag@hu.edu.et Fecal Microbial Communities. 13 Faculty of Medicine, Lund University, 221 00 Lund, Sweden; linda.kvist@outlook.com Microorganisms 2021, 9, 1153. 14 Department of Nutrition and Food Science, University of Ghana, Accra 00233, Ghana; geotoo@ug.edu.gh https://doi.org/10.3390/ 15 Department of Microbiology and Biochemistry of Dairy Products, Instituto de Productos Lácteos de microorganisms9061153 Asturias (IPLA-CSIC), 33300 Villaviciosa, Spain; lorena.ruiz@ipla.csic.es 16 Instituto de Investigación Sanitaria del Principado de Asturias (ISPA), 33011 Oviedo, Spain 17 Academic Editor: Pramod Gopal Department of Nutrition and Food Science, Complutense University of Madrid, 28040 Madrid, Spain; jmrodrig@vet.ucm.es 18 Nutrition Research Institute, Lima 15023, Peru; rpareja@iin.sld.pe Received: 3 May 2021 * Correspondence: rmpace@uidaho.edu (R.M.P.); smcguire@uidaho.edu (M.K.M.) Accepted: 24 May 2021 Published: 27 May 2021 Abstract: Previously published data from our group and others demonstrate that human milk oligosaccharide (HMOs), as well as milk and infant fecal microbial profiles, vary by geography. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in However, little is known about the geographical variation of other milk-borne factors, such as lactose published maps and institutional affil- and protein, as well as the associations among these factors and microbial community structures in milk iations. and infant feces. Here, we characterized and contrasted concentrations of milk-borne lactose, protein, and HMOs, and examined their associations with milk and infant fecal microbiomes in samples collected in 11 geographically diverse sites. Although geographical site was strongly associated with milk and infant fecal microbiomes, both sample types assorted into a smaller number of community Copyright: © 2021 by the authors. state types based on shared microbial profiles. Similar to HMOs, concentrations of lactose and Licensee MDPI, Basel, Switzerland. protein also varied by geography. Concentrations of HMOs, lactose, and protein were associated This article is an open access article with differences in the microbial community structures of milk and infant feces and in the abundance distributed under the terms and of specific taxa. Taken together, these data suggest that the composition of human milk, even conditions of the Creative Commons when produced by relatively healthy women, differs based on geographical boundaries and that Attribution (CC BY) license (https:// concentrations of HMOs, lactose, and protein in milk are related to variation in milk and infant fecal creativecommons.org/licenses/by/ microbial communities. 4.0/). Microorganisms 2021, 9, 1153. https://doi.org/10.3390/microorganisms9061153 https://www.mdpi.com/journal/microorganisms Microorganisms 2021, 9, 1153 2 of 17 Keywords: bacteria; breastmilk; gastrointestinal tract; HMO; human milk; infant; lactose; micro- biome; oligosaccharides; protein 1. Introduction Human milk is a complex biological fluid that provides all the nutritional require- ments that support infant growth and development. This is, in part, attributed to the fact that milk is a rich source of lactose, lipids, human milk oligosaccharides (HMOs), protein, and numerous other micronutrients [1]. Additionally, both culture-dependent and culture-independent methods have demonstrated the presence of microbiota in milk [2–6], with emerging data suggesting that these microbiota may play a role in seeding or supple- menting the nascent infant gastrointestinal (GI) microbiome [7]. In addition to supporting infant development, milk constituents (including lactose, protein, and HMOs) both directly and indirectly modulate host-associated microbial com- munities. As the principal carbohydrate source in milk, lactose is generally digested in the small intestine via lactase. Undigested lactose that reaches the large intestine is readily metabolized, and occasionally preferred over glucose, by resident microbes, in- cluding Lactobacillus and Bifidobacterium, into short-chain fatty acids and other volatile compounds [8–11]. Similarly, while most proteins are completely digested in the small intestine [1], partially digested proteins reaching the large intestine may be utilized by microbes [12]; although this is understudied in infants. In addition, some proteins (e.g., lactoferrin and secretory immunoglobulin A) function as host defense agents, modulating bacterial composition in the infant’s GI tract by repressing growth of pathogens. In contrast to lactose and protein, HMOs largely pass through the GI tract intact, as infants lack the enzymes to digest them [13,14]. Upon reaching the large intestine HMOs function principally as substrates for host-associated microbiota. However, HMOs not only promote the growth of microbes that are generally considered beneficial (e.g., Bifidobacterium [15]), they also function as antimicrobials that protect against pathogens, all of which contribute to shaping the infant GI microbiome and, in turn, infant health [16–21]. Results from the INSPIRE study (a large geographically and socioculturally diverse cohort) have previously demonstrated that the profiles of milk-borne immune factors [22], HMOs [23], and maternal and infant microbiomes [24,25] vary substantially across geo- graphical/sociocultural boundaries. As HMOs and other components of milk, including lactose and protein, are able to shape microbial abundance, we hypothesized that variation in these milk factors could be related to differences in the structure of milk and infant fecal microbial communities, as well as to the abundance of specific bacterial taxa. To test this hypothesis, we investigated relationships between and among microbial communities, and the concentrations of milk lactose, protein, and HMOs in milk and infant fecal samples collected from maternal–infant dyads in the INSPIRE study. 2. Materials and Methods 2.1. Study Design The participants in this study were recruited as part of the INSPIRE study, which has been described in detail [22–25]. All study procedures were approved by the Washington State University Institutional Review Board (#13264) and at each study location. Sample collection took place between May 2014 and April 2016 and was carried out as a cross- sectional, epidemiological, multi-cohort study. Briefly, samples were collected from 11 populations, including two from Ethiopia (rural population, ETR; urban population, ETU); Kenya (KE), Ghana (GN), two from The Gambia (rural population, GBR; urban population, GBU), Peru (PE), Spain (SP), Sweden (SW), and two from the United States of America (California, USC; Washington/Idaho, USW). To be eligible to participate, women had to be nursing or pumping ≥5 times a day and be ≥18 years of age. Exclusion criteria included: (1) current indication of a breast infection or breast pain that the woman did not consider Microorganisms 2021, 9, 1153 3 of 17 normal for lactation; (2) illness (i.e., self-reported fever, vomiting, severe cough, or diarrhea) in the last 7 days; and/or (3) antibiotic use in the previous 30 days. For inclusion, infants had to be described as healthy by their mothers, have no signs of acute illness (i.e., fever, vomiting, severe cough, diarrhea, or rapid breathing) in the previous 7 days, and have not received antibiotics in the previous 30 days. 2.2. Milk and Infant Fecal Sampling A total of 412 milk and 406 infant fecal samples were collected as part of the INSPIRE cohort. Descriptions of the sampling protocols for milk and feces have been previously described [24]. Briefly, milk was collected using gloved hands by participants or research personnel, after twice cleaning the breast with prepackaged castile soap towelettes (PDI, Inc, Woodcliff Lake, NJ, USA). Milk samples were collected via electric pump (Symphony, Medela Inc., Switzerland; PE, SW, USC, USW) or hand expression (ETR, ETU, KE, GN, GBR, GBU, SP) into sterile containers. Collected milk was immediately frozen (−20 ◦C), except in ETR where it was preserved in a 1:1 ratio with Milk Preservation Solution (Norgen Biotek, Ontario, CA) and frozen within 6 days. Approximately 1 g of feces was collected from diapers (Parent’s Choice; Walmart, Bentonville, AR, USA) or directly from the infant’s skin using a sterile, single-use scoop (Sarstedt AG & Co., Nümbrecht, Germany). Fecal samples were then placed into the accompanying sterile polypropylene container and frozen at −20 ◦C within 30 min of collection. For fecal samples collected in ETR, RNAlater (Ambion, Austin, TX, USA) was added to each fecal sample in a ∼1:4 ratio (feces:preservative) and frozen within 6 days. Milk and fecal samples were shipped on dry ice to the University of Idaho, where they were immediately frozen at −20 ◦C. 2.3. DNA Extraction and 16S rRNA Gene Amplification/Sequencing DNA was extracted from milk and infant fecal samples as previously described [24]. Extracted DNA from milk and infant fecal samples was subjected to a dual-barcoded, two-step, 30-cycle polymerase chain reaction (PCR) to amplify the V1-V3 hypervariable region of the 16S rRNA gene. In the first step, a 7-fold degenerate forward primer tar- geting nucleotide position 27 [26] and a reverse primer targeting nucleotide position 534 (positions numbered according to the Escherichia coli 16S rRNA gene) were used as de- scribed previously [27]. Amplicons were pooled to contain 50 ng of DNA from each sample. Size selection of amplicon pools were performed using AMPure beads (Beckman Coulter, Indianapolis, IN, USA), quality checked on a Fragment Analyzer (Advanced Ana- lytical Technologies, Inc., Ankeny, IA, USA), and quantified using the KAPA Biosciences Illumina library quantification kit and Applied Biosystems StepOne Plus real-time PCR system. Amplicons passing quality control for milk and feces were sequenced by sample type on separate MiSeq (Illumina, San Diego, CA, USA) sequencing runs (v3 paired-end, 300-bp protocol for 600 cycles at the University of Idaho Institute for Bioinformatics and Evolutionary Studies Genomics Core). 2.4. 16S rRNA Gene Amplicon Data Processing Samples were processed as previously described [24], with the following modifications. The DADA2-silva-derived taxonomy was edited to replace “NA” classifications with the next highest classification (e.g., if an amplicon sequence variant, ASV, was unclassified at the genus level but was classified at the family level as “Prevotellaceae”, it was given a genus-level classification of “Family_Prevotellaceae”). The ASV table was then filtered to remove ASVs assigned to mitochondria and chloroplasts. Furthermore, prevalence-based filtering to identify and remove potentially confounding sequences from milk samples that may have inadvertently arisen through sample collection, preparation, and reagent contamination [28,29], despite exercising aseptic technique [24], was performed using the R package decontam (v. 0.99.1) [30]. DNA extraction blanks (n = 23), generated and treated in parallel with milk samples, were used as negative controls in the filtering. As ETR milk samples were collected and stored using a preservative, these samples (n = 40) and their Microorganisms 2021, 9, 1153 4 of 17 respective negative controls (n = 5) were processed through decontam separately from the remaining cohort samples (n = 390) and their respective negative controls (n = 18). ASV tables were used as the input for the isNotContaminant function using the default parameters (method = “prevalence”, threshold = 0.5). After removing negative controls and ASVs lacking statistical support, ASV tables were merged in phyloseq and samples with less than 1000 reads (and their respective paired milk or infant fecal sample) were removed. An overview of the relative abundances of the most abundant genera before and after decontam filtering is presented in Figure S1 and a R markdown file containing code for decontam is included as a supplemental file. 2.5. Microbial Community State Type Analysis The R package DirichletMultinomial (v. 1.20.0) [31] was used to describe variability in microbiome data and cluster samples into community state types (i.e., “microbial lacto- types” for milk samples and “enterotypes” for infant fecal samples) based on the genus level abundance tables (filtered to remove genera present across all samples with a relative abundance of less than 0.01%). Model fit was determined based on the minimum Laplace goodness of fit. 2.6. Microbial Alpha and Beta Diversity For alpha and beta diversity analyses, samples were rarefied to 95% of the minimum sample read count. The rarefied data were used to generate the Bray–Curtis dissimilarity distance matrix, or converted to binary counts to generate the binary Jaccard distance matrix, and used for multidimensional scaling (MDS) and non-metric multidimensional scaling (NMDS). For MDS, the vegan adonis function was used to perform permutational multivariate analysis of variance (PERMANOVA) with 999 permutations to test for differ- ences between groups. The vegan envfit function was used to fit and determine goodness of fit and p-values for selected maternal and environmental factors (e.g., maternal body mass index (BMI), mode of delivery, and HMOs concentration) onto the Bray–Curtis NMDS ordination data using 9999 permutations. 2.7. Milk Composition Analysis Descriptions of the processing and analysis protocols for milk composition, includ- ing HMOs, have been previously described in detail [23]. HMOs quantified included: 2′-fucosyllactose (2′FL), 3-fucosyllactose (3FL), lacto-N-neotetraose (LNnT), 3′-sialyllactose (3′SL), difucosyllactose (DFlac), 6′-sialyllactose (6′SL), lacto-N-tetraose (LNT), lacto-N- fucopentaose (LNFP) I, LNFP II, LNFP III, sialyl-LNT (LST) b, LSTc, difucosyllacto- LNT (DFLNT), lacto-N-hexaose (LNH), disialyllacto-N-tetraose (DSLNT), fucosyllacto-N- hexaose (FLNH), difucosyllacto-N-hexaose (DFLNH), fucodisialyllacto-lacto-N-hexaose (FDSLNH) and disialyllacto-N-hexaose (DSLNH). HMOs were also grouped for analyses based on structural features and ratios: small HMOs (2′FL, 3FL, 3′SL, 6′SL, and DFLac), modified lactose (small HMOs and lactose), type 1 HMOs (LNT, LNFP I, LNFP II, LSTb, and DSLNT), type 2 HMOs (LNnT, LNFP III, and LSTc), α-1-2-fucosylated HMOs (2′FL and LNFP I), terminal α-2-6-sialylated HMOs (6′SL and LSTc), internal α-2-6-sialylated HMOs (DSLNT and LSTb), terminal α-2-3-sialylated HMOs (3′SL and DSLNT), ratio of HMO-bound sialic acid (Sia) to total HMOs (HMO-bound Sia/total HMOs), ratio of HMO- bound fucose (Fuc) to total HMOs (HMO-bound Fuc/total HMOs), and ratio of the ratio of HMO-bound Fuc to HMO-bound Sia (HMO-bound Fuc/HMO-bound Sia). Maternal secre- tor status phenotype was determined based on 2′FL presence or near absence (secretors, ≥200 nmol/mL; non-secretors, <200 nmol/mL) in the milk. Lactose concentrations were characterized using spectrophotometric assays as described previously [21]. Protein con- centrations were characterized using the Pierce bicinchoninic acid (BCA) assay (Cat#23225, Thermo Scientific, Waltham, MA, USA) using whole milk, diluted 1:20 with nanopure water, following manufacturer recommendations. Microorganisms 2021, 9, 1153 5 of 17 2.8. Statistical Analysis All statistical analyses were performed using R (v. 3.4.3 or v. 3.6.1). p Values were calculated using the Kruskal–Wallis test (followed by Dunn’s post hoc test, when applica- ble), Wilcoxon rank test, and Chi-squared test where appropriate. Infant weight-for-length z-scores were calculated using the R package zscorer (v. 0.3.1) [32]. The R packages vegan (v. 2.5-2) [33], phyloseq (v. 1.23.1) [34], ggplot2 (v. 3.0.0) [35], and pheatmap (v. 1.0.8) were used to perform and/or visualize alpha and beta diversity analyses, ordinations and cluster analyses. Taxonomic indicator values (IndVal) were calculated using the labdsv (v. 1.8-0) R package [36]. The R packages Hmisc (v. 4.4.2) [37] and corrplot (v. 0.84) [38] were used to perform and project Spearman correlations, respectively, among bacterial genera (relative abundance ≥1% within sample types), lactose, protein, and HMOs. Where applicable, p values were FDR corrected with the R p.adjust function. Values are given as mean ± standard deviation (SD), unless otherwise indicated. Statistical significance was declared at p < 0.05 and/or FDR p ≤ 0.1. 3. Results 3.1. Cohort Demographics Data for 357 maternal–infant dyads were available for analyses after microbiome sequence processing and quality control. On average, maternal age was 27.4 ± 6.1 years, with milk and infant fecal samples collected an average of 64.6 ± 21.9 days postpartum. Overall, the majority (86%) of births were via vaginal delivery, and the frequency of exclusive breastfeeding at the time of sample collection was 60%. Consistent with prior reports of the INSPIRE cohort [22–25], there were myriad differences in demographics across populations. Additional selected demographics for these dyads are detailed in Table S1. 3.2. Milk and Infant Fecal Microbiomes We observed a total of 5303 ASVs, of which 2085 were observed in milk (14,525 ± 15,479 average reads) and 3935 were observed in infant fecal samples (11,544 ± 6029 av- erage reads), respectively. These ASVs corresponded to 22 phyla and 486 genera, with milk containing 21 phyla and 363 genera and infant feces containing 12 phyla and 272 genera. Principal coordinates analysis confirmed that milk and infant fecal microbiomes differed with respect to community structure and membership (Figure S2). Milk samples were dominated by Staphylococcus (28%, mean relative abundance) and Streptococcus (26%), followed by Corynebacterium (6%), Propionibacterium (Cutibacterium; 5%), unclassified Xan- thomonadaceae (3%), and Lactobacillus (3%) (Figure S2C). Two thirds of the overall relative abundance of microbiota in infant feces was attributed to five genera: Streptococcus (17%), Escherichia/Shigella (16%), Bifidobacterium (12%), Veillonella (12%), and Bacteroides (10%) (Figure S2C). Community state type (CST, communities of similar microbial composition and abun- dance) analysis has been used to explore variation of the microbial communities of the feces (i.e., enterotypes) [39–42], vagina [43], and milk [3]. To identify and examine CSTs, we applied Dirichlet multinomial mixtures modelling to both milk and infant fecal micro- biomes. Milk samples formed four clusters or microbial lactotypes (i.e., L1 through L4) (Figure 1A), whereas infant fecal samples formed two clusters or microbial enterotypes (i.e., E1 and E2) (Figure 1B). Both microbial lactotypes and enterotypes differed with re- spect to parity, maternal age, maternal BMI, and exclusive breastfeeding status (Tables S2 and S3). Lactotypes also differed by time postpartum and maternal secretor status. The distribution of populations within milk and infant fecal CSTs varied. Among lactotypes, L4 was comprised exclusively of rural Ethiopian (ETR) subjects, although not all ETR subjects belonged to the L4 cluster (Figure 1A). L1 was mainly comprised of individuals from the Americas and Europe (i.e., PE, SP, SW, USC, and USW), while L2 and L3 were both largely comprised of individuals from Africa (i.e., L2-ETU, GBR, GBU, and KE; L3-GN), although L3 contained a marginal proportion of participants from the SP cohort. Similarly, Microorganisms 2021, 9, x FOR PEER REVIEW 6 of 17 L4 was comprised exclusively of rural Ethiopian (ETR) subjects, although not all ETR sub- jects belonged to the L4 cluster (Figure 1A). L1 was mainly comprised of individuals from Microorganisms 2021, 9, 1153 the Americas and Europe (i.e., PE, SP, SW, USC, and USW), while L2 and L3 were b6 otfh1 7 largely comprised of individuals from Africa (i.e., L2-ETU, GBR, GBU, and KE; L3-GN), although L3 contained a marginal proportion of participants from the SP cohort. Similarly, infant enterotype membership varied largely by African and non-African populations, inwfiathn tinednitveirdoutyaplse frmomem thbee rAshmiperivcaarsi eadndl aErugreolypeb gyroAufpriicnagn maanidnlnyo wni-tAh fEri1c (aFnigpuorpe u1lBa)t.i Aonns , waistshesisnmdievnitd oufa tlhsef rroemlattihoensAhmipe breictawseaennd inEfuanrot pfeecgarl oeunpteirnogtympaeisn alyndw tihthe lEa1ct(oFtiygpuerse o1fB t)h.eAirn arsessepsescmtievnet mofotthheerrse lraetvioenaslehdip mboettwheeresn winitfha nmt filekc atlheant tbeerolotnypgeeds atnod Lt1h eorl aLc3to htyapde sa olafrtgheeri r rpesroppeoctritvioenm oof tihnfearnstrse tvheaat lbeedlomngoethde tros Ew1i t(h67m%i alkndth 6a3t%b,e rleosnpgeecdtivtoelyL)1; morotLh3erhsa wditahl mariglke r ptrhoapt obretlioonngoefdi tnof aLn2t hs athda at lbaergloenr gperdoptoorEti1on(6 o7f% inafnandts6 3th%a,t rbeeslpoencgteivde tloy )E; 2m (o60th%e)r.s Hwoiwthevmeirl,k tihnaftabnetslo onfg medottoheLr2s hwaidtha mlairlgke rthparto bpeolrotniognedo ftion fLa4n tws tehrea tsbpelliot nbgeetwd etoenE 2E1(6 (04%9%). )H aonwde Eve2r , in(5f1a%nt)s. of mothers with milk that belonged to L4 were split between E1 (49%) and E2 (51%). TThheea abbuunnddaanncceea nanddp rperveavlaelnecnecoe fosfe vseevraelrgale ngeernaewrae rweearses oacsisaotceidatwedit hwiinthd ivinidduivaildmuaillk amndilkin afnadnt infefcaanlt CfeScTasl bCaSsTesd boanseindd oinca itnodriscpateocri esspaenciaelsy sainsa(lFyisgius r(eFi1g;uFrDe R1; pFD4%, each), with an additional seven and eight HMOs also associated with milk and infant fecal microbiomes, respectively. Examination of HMOs by shared characteristics revealed that the concentration of HMO-bound fucose explained the most variance within the milk microbiome (~7%), and the ratio of HMO-bound fucose to HMO-bound sialic acid explained the most variance within the infant fecal microbiome (~12%). Thus, variation in milk and infant fecal microbiomes were related to differences in maternal and infant characteristics, as well as the concentrations of milk-borne factors. MMicriocororogragnainsmissm 2s022012, 19,, 9x, F11O5R3 PEER REVIEW 9 o9f o1f71 7 FiFgiugruer e3.3 A. Asssoscoiacitaiotinosn bsebtewtweeene nmmataetrenranla/iln/fiannfat ncthcahraacrtaecrtiesrtiisctsi,c ms, imlki lfkacftaocrtso,r as,nadn md imcriocrboiabli aclocmommunuintyit sytrsutrcutuctruerse osfo mf milkil k anadn dinifnafnatn ftefceecse. sE. nEvnfvitfi wt wasa ussuesde dtot ommodoedle ml mataetrenranla alnadn dinifnafnatn cthcahraarcatcetreisrtiisctsic (sp(ipnikn)k, )t,hteh ecocnocnecnetnratrtaiotino nofo mf milkil klalcatcotsoes e (li(glihgth gtrgereene)n, p),rpotreoitne i(nda(drka rgkregerne)e,n a)n, da nHdMHOMs O(insd(invdidivuiadl uHaMl HOMs iOn sliignhtli bglhuteb, lHuMe, OH MgrOougprionugsp ing dsairnk dbalurke) balguaei)nastg (aAin)s t m(iAlk) amndilk (Ba)n idnf(aBn)t ifnefcaanl tmfeiccraolbmiailc crobmiamlucnoimtym sutrnuicttyusretr u(Bcrtauyre−C(Burratiys–, CNuMrtDisS, ;N n M= D34S0;)n. F=ea3t4u0r)e.s Faeloantugr tehsea ylo-anxgesth ien yb-oatxhe s painneblso tahrep oanrdeelsreadre boarsdeder oedn bthaes eRd2 ovnaltuhees Rfr2ovma ltuhees mfriolkm atnhaelymsiisl,k saonrtaeldy swisi,tshoinrt emdawteirtnhailn/imnfaatnetr nchaal/rainctfearnitstcihcsa raancdte mrisiltkic s faacntodr gmriolukpfiancgtos rfrgormou hpiginhgesstf troo mlowheigsht R 2 est. StoignloifwiceasnttR fe2a. tSuirgens i(fiFcDaRn tpf ≤e a0t.u1)r easre( FdDenRotped≤ w0i.t1h) aanre asdteenriostke.d 2′w-fuitchosaynllaasctteorsies k. (22′F′-Lfu),c 3o-sfyulcloacstyolslaec(t2o′sFeL ()3, F3L-f)u, cloascytoll-aNct-onseeot(e3tFrLa)o,slea c(tLoN-Nn-Tn)e, o3t′e-stiraaloyslela(cLtoNsneT ()3,′S3L′-)s,i adliyflulaccotsoyslela(c3t′oSsLe) ,(DdiFfulacco)s, y6l′l-asciatolyslela(DctFolsaec ), (6′S′L), lacto-N-tetra′ose (LNT), lacto-N-fucopentaose (LNFP) I, LNFP II, LNFP III, sialyl-LNT (LST) b, LSTc, difucosyllacto- LN6 T-s (iDalFyLllaNcTto),s lea(c6toS-LN)-,hlaecxtaoo-Nse- (tLetNraHo)s,e d(iLsNialTy)l,lalaccttoo--NN--tfeutrcaoopseen (tDaoSsLeN(LTN), FfuPc)oIs,yLlNlacFtPo-INI, -LhNexFaPoIsIeI ,(FsiLaNlyHl-L),N dTifu(LcoSsTy)lbla,cLtoS-Tc, Nd-hifeuxcaoossyel l(aDctFoL-LNNHT), (fDuFcLodNiTsi)a,llyalcltaoc-tNo--lhacetxoa-oNse-h(eLxNaoHs)e, d(FisDiaSlLyNllaHct)o a-Nnd-t edtirsaiaolsyell(aDcStoL-NNT-h),efxuacoossey (llDacStLoN-NH-)h; esxmaoaslle H(FMLNOHs ), (2d′FiLfu, c3oFsLy,l 3la′ScLto, -6N′S-Lh,e axnados DeF(DLaFcL)N, mHo)d, fifuiecodd laiscitaolsyell (ascmtoa-llla HctoM-NO-sh aenxda olascet(oFsDe)S, LtyNpHe 1) aHnMd Odiss i(aLlNylTla, cLtNo-FNP- hI,e LxNaoFsPe I(ID, LSLSTNbH, ); ansdm DalSl LHNMTO),s t(y2p′Fe L2, 3HFML,O3s′S (LL,N6′nSTL,, LanNdFDP FILIIa, ca)n, dm oLdSiTfice)d, αla-1ct-o2-sfeu(csomsyallal tHedM HOMs aOnsd (l2a′cFtLo saen),dt yLpNeF1PH IM), Otesrm(LiNnaTl, αL-N2-F6P- I, siaLlNylFaPteIdI, HLSMTbO,sa n(6d′SDLS aLnNdT L),StyTpc)e, 2inHteMrnOasl (αL-N2-n6T-s, iLaNlyFlaPteIIdI ,HanMdOLsS T(cD),SαL-N1-T2 -afuncdo sLySlTatbe)d, tHerMmOinsa(l2 ′αF-L2-a3n-sdiaLlNylFaPteId), HteMrmOinsa l (3α′S-L2 -a6n-sdia DlySlaLtNedT)H, rMatOios o(6f ′HSLMaOn-dboLuSnTdc) ,siianltiecr ancaildα (-S2i-a6)- stioa ltyoltatle Hd MHOMsO (sH(MDOSL-bNoTunadn dsiLalSiTc ba)c,idte/rtomtainl aHl Mα-O2-s3)-, sriatliyol aotef d HHMMOO-bso(u3n′SdL fuancodsDe S(FLuNcT) )t,or taotitoalo Hf HMMOOs (-HboMunOd-bsoiaulincda cfuidco(Sseia/t)ottoalt oHtaMl HOMs),O asn(dH rMatOio- boof uthned rsaiatiloic oafc HidM/tOot-abloHuMndO Fsu),cr atoti o HoMf OH-MboOu-nbdou Sniad (fHucMosOe-(bFouucn)dto futoctoasleH/HMMOOs -(bHoMunOd- bsoiaulnicd afcuidco).s e/total HMOs), and ratio of the ratio of HMO-bound Fuc to HMO-bound Sia (HMO-bound fucose/HMO-bound sialic acid). Microorganisms 2021, 9, 1153 10 of 17 3.6. Correlation of Milk Factors with Bacterial Taxa Abundance A correlation analysis was performed to identify associations between and among concentrations of milk lactose, protein, HMOs, and the relative abundances of specific bac- terial genera in milk and infant feces. Similar to prior observations of the milk microbiome and HMOs [19], stronger correlations were observed within a class of constituents (e.g., HMO-to-HMO and bacterium-to-bacterium) than between milk macronutrients/HMO and bacterial taxa (Figure 4). This was mainly attributed to the grouping of HMOs with shared features such as α-1-2-fucosylated HMO (e.g., 2′FL and LNFP I, $ = 0.74) or sialylated HMOs (e.g., 6′SL and LSTc, $ = 0.65). Significant bacterium-to-bacterium correlations were also observed within and among milk and infant fecal samples. With respect to bacteria, the strongest inverse associations were observed between Bacteroides and Streptococcus in infant feces ($ = −0.40), Dyella and Rhizobium in milk ($ = −0.38), and Gemella in milk and Lactobacillus in infant feces ($ = −0.29). The strongest positive associations were observed between Dyella and unclassified Xanthomonadaceae in milk ($ = 0.68), Corynebacterium 1 and Kocuria in milk ($ = 0.43), Bacteroides and Parabacteroides in infant feces ($ = 0.42), and Bifidobacterium and Kocuria in milk ($ = 0.41). Interestingly, while the relative abundances of Lactobacillus were positively correlated between infant fecal and milk samples ($ = 0.36), the relative abundances of Bifidobacteria in feces and milk were inversely correlated ($ = −0.18). Associations were also observed between milk constituents and both milk and infant fecal microbiota (Figure 4). For example, the relative abundances of milk and infant fecal Lactobacillus were inversely related to concentrations of fucosylated HMOs, whereas relative abundances of milk and infant fecal Veillonella were inversely correlated with concentrations of sialylated HMOs. Given the specific adaptations for HMO utilization of specific Bifidobacteria species, it is noteworthy that the abundance of infant fecal Bifidobacteria was only significantly associated with the concentration of fucosylated HMOs, except for an inverse correlation with DFLNH; however, the relative abundance of milk Bifidobacteria was inversely related to the concentration of several fucosylated HMO, including 2′FL. Positive correlations were observed between milk and infant fecal Bifidobacteria with DSLNT, DSLNH, and other sialylated HMOs. In contrast, Streptococcus in milk (but not infant feces) was positively associated with fucosylated HMOs. Correlations between the relative abundance of bacteria in milk and infant feces and the concentrations of protein and lactose were few and weak; although, in general, correlations between bacteria and lactose tended to be stronger than those between bacteria and protein. Taken together, although within class associations of milk-borne factors were stronger, numerous associations between these same factors and specific milk and infant fecal microbiota were identified. MicrMooircgroaonrigsamnsis2m0s2 210, 291, ,1 91,5 x3 FOR PEER REVIEW 11 o1f1 1o7f 17 FiguFirgeu4r.e C4.o Crroerlraetliaotniosnasm amonogngm micircorboibailalt atxaxaaa annddm miliklk ffaaccttoorrss.. SSppeeaarrmaann rraannkk ccoorrrerlealtaitoinosn samamonogn gmmilki l(kM()M a)nadn idnfiannfta nt fecaflec(IaFl )(ImF)i cmroicbrioablitaalx taa,xlaa, clatocstoes,ep, rportoetieni,nH, HMMOOs,s,a annddH HMMOO ggrroouuppiinnggss aarree sshhoowwnn ((nn == 33404 0ppaiarierde ddydaydasd).s O). nOlyn lcyorcroerlraetiloantiso ns with FDR≤ p ≤ 0.1 are shown. The size of the circle represents the magnitude of the correlation, and the color represents the withdiFreDcRtiopn of0 t.h1ea croersrheloawtionn. :T rhede s(nizeegoatfivthee) acnirdc lbelrueep (rpeosseintitvset)h. eFemataugrneist uwdeereo hf itehrearccohrirceallalyti ocnlu,satnerdedth beacsoedlo ornre Sppreeasremntasnt he direccotriroenlaotifotnhse. Fceoartruerlaet ciolans:serse dar(en deegnaotitveed) bayn tdheb lcuoelo(rpeods sitqiuvaer)e. sF neaextut rteos thwee freeathuireer anracmhiec aolnly thcelu lesftte,r aesd foblalosewds:o lnigShpt ebalurme, an corrinedlaitviiodnusa.l FHeMatOurse; dcalarkss besluaer, eHdMenOo gterdoubpyintghse; lcigohlot rgerdeesnq,u laarcetossne;e xdtartko gthreeefne, apturorteeina; meilko ngetnheerale, fpt,inaks; fionlflaonwt sfe: clailg ht bluge,eninedrai,v piduurpaleH. 2M′-fOucso; sdyallrakctbolsuee (,2H′FLM),O 3-gfurocouspyilnlagcst;olsieg (h3tFgLr)e, elanc,tola-Nct-onseeo;tdetarrakosger e(LeNn,npTr)o, 3te′-isnia; lmylillakctgoesen e(3r′aS,Lp),i ndkif;uicnof-ant fecaslyglleacnteorsae, (pDuFrlpacle),. 62′-′s-ifaulcyollsaycltloascet o(6s′eSL(2),′ lFaLct)o, -3N-f-utectorsayolslea (cLtoNsTe)(, 3laFcLto),-Nla-cftuoc-oNp-ennetoaotester (aLoNseFP(L) NI, nLTN)F, P3 ′I-Is, iLaNlyFllPa cIItIo, sseia(l3y′lS- L), difuLcNoTsy (lLlaScTt)o bs,e L(SDTFc,l adci)f,u6co′-ssyialllaycltloa-cLtNosTe ((D6F′SLLN),Tl)a, clatoct-oN-N-te-htreaxoasoese( L(LNNTH),),l adcistoia-lNyl-lfauctcoo-pNe-ntettaroasoese( L(DNSFLPN)TI,),L fuNcFoPsyIllIa, cLtoN-FP N-hexaose (FLNH), difucosyllacto-N-hexaose (DFLNH), fucodisialyllacto-lacto-N-hexaose (FDSLNH) and disialyllacto- III, Nsi-ahleyxl-aLoNseT (D(LSLSTN)Hb),; sLmSTacll, HdMifuOcso (s2y′FllLa,c t3oF-LL,N 3′TSL(,D 6F′SLLN, aTn)d, lDacFtLoa-cN),- hmeoxdaiofiseed (lLaNctoHse), (dsmisaialll yHllMacOtos -aNn-dt eltarcatoossee),( DtySpLe N1 T), fucoHsMylOlasc t(oL-NNT-,h LeNxaFoPs eI, (LFNLNFPH I)I,, LdSifTubc,o asnydll aDcStoL-NNT-)h, etyxpaoe s2e H(DMFOLsN (HLN),nfTu,c LodNiFsPia lIyIIl,l aacntdo -LlaScTtco),- Nα--1h-e2x-fauocsoesy(FlaDteSdL HNMHO) as nd disi(a2l′FyLll aacntdo -LNN-hFePx Ia),o tseerm(DinSaLl NαH-2-)6; -ssmiaalylllaHteMd HOMs (O2′sF (L6,′S3LF aLn,d3 ′LSSLT,c6)′,S iLnt,earnadl Dα-F2L-6a-cs)i,almyloadteidfi eHdMlaOcst o(sDeS(LsNmTa lalnHd MLSOTsb)a, nd lacttoesrem),intaylp αe-21-3H-sMialOyslat(eLdN HT,MLONsF (P3′SI,LL aNndF PDISIL, NLSTT), br,atainod ofD HSLMNOT-b),outynpde s2ialHicM acOids (SLiNa)n tTo, tLoNtalF PHMIIIO, sa n(HdMLOST-bc)o,uαn-d1 -2- fucoSisay/ltaotteadl HMMOOs)s, (r2a′tFioL oafn HdMLON-FbPouI)n,dt efrumcoinsea l(Fαu-c2)- 6to-s tioatlayll aHteMdOHs M(HOMsO(6-b′SoLunadn dFuLcS/Ttoct)a, li nHtMerOnasl),α a-n2d-6 r-astiiaol yolfa ttheed rHatMioO s (DSoLfN HTMaOnd-bLoSuTnbd) F, tuecr mtoi HnaMl αO--2b-o3u-sniadl ySliaat (eHdMHOM-bOosu(n3d′S FLuacn/HdMDOSL-bNoTu)n,dra Stiiao).o f HMO-bound sialic acid (Sia) to total HMOs (HMO-bound Sia/total HMOs4)., rDatiisocuofssHioMnO -bound fucose (Fuc) to total HMOs (HMO-bound Fuc/total HMOs), and ratio of the ratio of HMO-bound Fuc to HMO-bound Sia (HMO-bound Fuc/HMO-bound Sia). Human milk contains a diversity of nutrients and bioactive factors known or postu- 4.laDteidsc tuos isnifolunence maternal and infant health. Here we tested the hypothesis that variation in the concentrations of milk lactose, protein, and HMOs would be associated with differ- Human milk contains a diversity of nutrients and bioactive factors known or postu- lated to influence maternal and infant health. Here we tested the hypothesis that variation in the concentrations of milk lactose, protein, and HMOs would be associated with dif- Microorganisms 2021, 9, 1153 12 of 17 ferences in the community structure and abundance of milk and infant fecal microbiota. Although milk and infant fecal microbiota are known to vary by geographical location, we found milk and infant fecal microbiota grouped into four microbial lactotypes and two enterotypes, respectively, based on similarities in community composition. Interestingly, while the CSTs within milk and infant feces contained representative genera (e.g., Lacto- bacillus in L3 and Bacteroides in E1), several genera were shared among all samples; for example, Staphylococcus and Veillonella were core genera in milk and infant fecal samples, respectively. Interestingly, despite Staphylococcus spp. (e.g., Staphylococcus aureus) being commonly implicated as a common etiological agent of mastitis [45,46], mastitis was not reported by any of the women in the current study. Indeed, Staphylococcus spp. are common constituents of milk microbial communities [3,47,48]. While differences in traits and virulence factors among strains of staphylococci [49] may help to partially explain why not all carriers of Staphylococcus spp. go on to develop mastitis, it also indicates that the presence of Staphylococcus spp. in milk alone is not sufficient to treat with antibiotics, and that more research related to the microbial etiology of mastitis is needed. Similar to HMOs, concentrations of lactose and protein displayed a substantial amount of interindividual variation that differed among population cohorts and CSTs. Concentra- tions of HMOs are largely determined by genetic factors [50]. For example, individuals with a functioning FUT2-encoded fucosyltransferase (i.e., secretors) are able to synthe- size α-1-2-fucosylated HMOs such as 2′-fucosyllactose (2′-FL) and lacto-N-fucopentaose I (LNFP I), whereas individuals lacking a functional FUT2 allele (i.e., non-secretors) are unable to synthesize these glycans [51]. Less is known about the underlying genetics that contribute to the concentrations of milk lactose and protein. There are some data that suggest host genetics may play a role in milk lactose concentrations; e.g., lactose concentrations differ among ABH and Lewis secretor types [52], and differences in lactose concentrations have been observed in milk produced by women living in five different countries using metabolomics [53]. In contrast, while single nucleotide polymorphisms impacting bioactivity of some milk proteins have been identified [54], none have been strongly related to differences in milk protein concentration [55]. Regional differences in maternal nutrition may explain some of the variation in concentrations that we observed, although this is unlikely to be a significant contributor as both milk lactose and protein levels are relatively unaltered by diet [56,57]. The full extent of the impact of host genetics on lactose and total protein concentrations remains to be determined. The variation in protein and lactose concentrations among population cohorts may also have implications for current recommendations for dietary consumption of these nutrients. Adequate intake (AI) levels of nutrients for infants from 0 to 6 months are estimated based on an exclusive human milk diet (average concentration of the nutrients in milk coupled with milk intake volume of 0.78 L/d) by healthy, full-term infants born to healthy, well-nourished mothers [58]. For example, the AIs for carbohydrates (in human milk being almost exclusively lactose) and protein are 74 g/L and 11.7 g/L, respectively [58]. In the present study, 70 percent of overall participants produced milk that met or exceeded the value used to establish the AI for carbohydrates/lactose. However, this varied by cohort, ranging from 47 percent of USW women to 100 percent of USC women. Similarly, whereas 92 percent of overall participants produced milk with protein concentrations that met or exceeded the value used to establish the AI for protein, this ranged from 66 percent of USW women to 100 percent of GBR and PE women. Although in the current study we did not assess milk intake volume of the infants, the differences in concentrations of lactose and protein across cohorts suggest that average consumption of these important macronutrients may differ by population. The complex microbial communities present in milk and infant feces were associ- ated with numerous maternal/infant factors as well as milk HMO, lactose, and protein concentrations. Not surprisingly, based on prior data [24], population cohort explained the most variance within both microbiomes; however, maternal age, parity, and exclusive Microorganisms 2021, 9, 1153 13 of 17 breastfeeding were also significantly associated with variation. Interestingly, microbial lac- totype was significantly associated with variation in the infant fecal microbiome, in support of the hypothesis that milk microbiota contribute to and influence infant GI microbiome composition [4,7,59]. While the proportion of secretors differed among microbial lactotypes, maternal secretor status per se was not associated with the microbial community structure of milk or infant fecal microbiota, consistent with findings from the CHILD and TwinsUK study cohorts [19,60]. Instead, microbial community structures of milk and infant fecal samples were associated with concentrations of α-1-2-fucosylated HMOs and HMO-bound fucose. This is likely related to underlying host genetics that results in a large degree of variation in the concentrations of α-1-2-fucosylated HMOs in the milk produced by secretors [23]. Although we did not analyze dietary patterns in this study, maternal diet may play a role in altered patterns of α-1-2-fucosylated HMO composition [61]. Several other HMOs (e.g., 3FL, LNFP III, LSTb, and DSLNT) were also found to be related to the microbial community structures of milk and infant feces. Interestingly, DSLNT, hypothesized to protect against necrotizing enterocolitis [62–64], was among the top three HMOs that explained the most variance in the structure of both milk and infant fecal microbiomes. In a correlation analysis, DSLNT was also positively associated with the abundance of Bifidobacterium in both milk and infant feces; Bifidobacterium is considered to be health-promoting during infancy [65] and has been found to positively correlate with DSLNT concentrations in milk [64]. It is notable that all the infants in this study were reported as born at term and currently healthy. While concentrations of lactose were associated with the microbial community struc- ture of milk, concentrations of protein were not. Conversely, while concentrations of protein were associated with the microbial community structure of infant feces, concentrations of lactose were not. These differences are likely reflective of host factors and in the microenvi- ronment of the mammary gland and the GI tract. For example, lactose plays a major role in controlling milk volume via maintenance of the osmolarity of milk in the mammary gland and is mostly metabolized prior to reaching the lower GI tract. Lactose concentrations were positively correlated with the abundance of Streptococcus in milk, a genus known to be capable of fermenting lactose. However, given the abundance of lactose in milk, it is unlikely to be a rate-limiting substrate for bacterial growth. Instead, lactose may function to modulate bacterial abundance by inducing metabolization of other milk-borne factors, similar to HMO-induced amino acid utilization [66]; however, this remains to be examined. Several limitations to the current study should be noted. As we only examined total protein, we were unable to examine associations among microbiota and individual proteins or classes of proteins, such as secretory immunoglobulin A and lactoferrin, both of which can directly influence microbial ecology [67,68]. Additionally, although we examined the associations between microbiota and concentration of 19 HMOs that represent the majority of HMOs present in milk, to date over 200 HMO species (most present in low abundance) have been identified in human milk [69]. However, the 19 analyzed HMOs not only represent the majority of all HMOs by mass, they also describe the entire known chemical space of HMOs, including type 1 and 2 structures, branching, as well as all types of fucosylation and sialylation. Whether the other more complex (structure redundancy) and less abundant HMOs have an impact on milk and infant fecal microbiota is unknown and needs additional research. Finally, HMO concentrations were only measured in milk and not infant feces; as such we could not examine how HMOs may have been metabolized as they pass through the infant GI tract and any relationships with resident microbiota. Additionally, while we attempted to standardize and/or optimize milk and fecal collection, handling, and analysis protocols, due to practical considerations this was not always possible. For example, lack of access to reliable refrigeration required us to mix and store all ETR samples with a preservative, which may have influenced downstream analyses. However, key strengths of this work are the large and globally diverse cohort of dyads; Microorganisms 2021, 9, 1153 14 of 17 recruitment of relatively healthy participants in each cohort; and inclusion of appropriate controls and filtering parameters applied prior to analysis. 5. Conclusions Taken together, our results demonstrate that variation in human milk and infant fecal microbial communities are associated with differences in the concentrations and profiles of milk lactose, protein, and HMOs. Future work should focus on understanding how these associations develop and mature over the course of lactation and infant development. Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/microorganisms9061153/s1, Figure S1: Relative abundances of the initial top 25 bacterial genera in milk before and after prevalence-based filtering; Figure S2: Milk and infant fecal microbiome beta diversities and most abundant taxa; Figure S3: Differences in the alpha and beta diversities of milk and infant fecal community state types; Figure S4: HMO composition; Table S1: Selected cohort characteristics; Table S2: Milk microbial community state type (lactotype) characteristics; Table S3: Infant fecal microbial community state type (enterotype) characteristics; Table S4: Indicator taxa of the microbial lactotypes; Table S5: Indicator taxa of the infant microbial enterotypes; Table S6: HMO concentrations and ratios among microbial lactotypes; Table S7: HMO concentrations and ratios between enterotypes; Supplementary file: Identification and removal of contaminant ASVs from milk samples R markdown file. Author Contributions: Conceptualization, R.M.P., J.E.W., C.L.M., J.A.F., D.W.S., A.M.P., L.J.K., J.M.R., R.G.P., M.A.M., L.B. and M.K.M.; data curation, R.M.P., J.E.W., B.R. and L.B.; formal analysis, R.M.P.; funding acquisition, C.L.M., J.A.F., L.R., J.M.R., M.A.M., L.B. and M.K.M.; investigation, R.M.P., J.E.W., M.A.M., L.B. and M.K.M.; methodology, R.M.P., J.E.W., B.R., M.A.M., L.B. and M.K.M.; project administration, M.A.M. and M.K.M.; resources, M.A.M., L.B. and M.K.M.; supervision, M.A.M. and M.K.M.; visualization, R.M.P.; writing—original draft, R.M.P.; writing—review and editing, R.M.P., J.E.W., B.R., K.A.L., C.L.M., W.J.P., J.A.F., E.W.K.-M., E.W.K., S.M., S.E.M., A.M.P., D.G.K., L.J.K., G.E.O., L.R., J.M.R., R.G.P., M.A.M., L.B. and M.K.M. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by the National Science Foundation IOS-BIO 1344288; the Ministry of Economy and Competitiveness, Spain (AGL2013-4190-P); the European Commission (624773-FP-7-PEOPLE-2013-IEF, to LR); and in part by National Institutes of Health NICHD R01- HD092297 and COBRE Phase III grant P30GM103324. Sterile, single-use milk collection kits were provided by Medela Inc. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Washington State University (#13264, July 2013). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Data and materials that support the findings of this study are available upon request from the corresponding authors. Acknowledgments: We thank Andrew Doel (Medical Research Council Unit, The Gambia) for field supervision and logistics planning and Alansan Sey for questionnaire administration and taking anthropometric measurements in The Gambia; Jane Odei (University of Ghana) for supervising field data collection in Ghana; Katherine Flores (Washington State University), Dubale Gebeyehu (Hawassa University), Haile Belachew (Hawassa University), and Birhanu Sintayehu for planning, logistics, recruiting, and data collection and the administration and staff at Adare Hospital in Hawassa for assistance with logistics in Ethiopia; Catherine O. Sarange (Egerton University) for field supervision and logistics planning and Milka W. Churuge and Minne M. Gachau for recruiting, questionnaire administration, and taking anthropometric measurements in Kenya; Gisella Barbagelatta (Instituto de Investigación Nutricional) for field supervision and logistics planning, Patricia Calderon (Instituto de Investigación Nutricional) for recruiting, questionnaire administration, and taking anthropometric measurements, and Roxana Barrutia (Instituto de Investigación Nutricional) for the management and shipping of samples in Peru; Leonides Fernandez, and Irene Espinosa (Complutense Univer- sity of Madrid) for technical assistance, expertise, and review of the manuscript and M Ángeles Microorganisms 2021, 9, 1153 15 of 17 Checa (Zaragoza, Spain), Katalina Legarra (Guernica, Spain), and Julia Mínguez (Huesca, Spain) for participation in the collection of samples in Spain; Kirsti Kaski and Maije Sjöstrand (both Hels- ingborg Hospital) for participation in the collection of samples, questionnaire administration, and anthropometric measurements in Sweden; Renee Bridge and Kara Sunderland (both University of California, San Diego); Janae Carrothers and Shelby Hix (Washington State University) for logistics planning, recruiting, questionnaire administration, sample collection, and taking anthropometric measurements in California and Washington; Jessica A. Zachek, Kelcey McBride, Elizabeth D. Benda, Romana J. Hyde, and Morgan M. Potton for their help in sample processing and/or analysis; and Glenn Miller (Washington State University) for his expertise and critical logistic help that were needed for shipping samples and supplies worldwide. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. References 1. Ballard, O.; Morrow, A.L. Human milk composition. Pediatr. Clin. N. Am. 2013, 60, 49–74. [CrossRef] [PubMed] 2. Hunt, K.M.; Foster, J.A.; Forney, L.J.; Schütte, U.M.E.; Beck, D.L.; Abdo, Z.; Fox, L.K.; Williams, J.E.; McGuire, M.K.; McGuire, M.A. Characterization of the diversity and temporal stability of bacterial communities in human milk. PLoS ONE 2011, 6, e21313. [CrossRef] 3. Moossavi, S.; Sepehri, S.; Robertson, B.; Bode, L.; Goruk, S.; Field, C.J.; Lix, L.M.; de Souza, R.J.; Becker, A.B.; Mandhane, P.J.; et al. Composition and variation of the human milk microbiota are influenced by maternal and early-life factors. Cell Host Microbe 2019, 25, 324–335. [CrossRef] [PubMed] 4. Martin, R.; Langa, S.; Reviriego, C.; Jimenez, E.; Marin, M.L.; Xaus, J.; Fernandez, L.; Rodriguez, J.M. Human milk is a source of lactic acid bacteria for the infant gut. J. Pediatr. 2003, 143, 754–758. [CrossRef] [PubMed] 5. Perez, P.F.; Doré, J.; Leclerc, M.; Levenez, F.; Benyacoub, J.; Serrant, P.; Segura-Roggero, I.; Schiffrin, E.J.; Donnet-Hughes, A. Bacterial imprinting of the neonatal immune system: Lessons from maternal cells? Pediatrics 2007, 119, e724–e732. [CrossRef] [PubMed] 6. Collado, M.C.; Delgado, S.; Maldonado, A.; Rodriguez, J.M. Assessment of the bacterial diversity of breast milk of healthy women by quantitative real-time PCR. Lett. Appl. Microbiol. 2009, 48, 523–528. [CrossRef] 7. Pannaraj, P.S.; Li, F.; Cerini, C.; Bender, J.M.; Yang, S.; Rollie, A.; Adisetiyo, H.; Zabih, S.; Lincez, P.J.; Bittinger, K.; et al. Association between breast milk bacterial communities and establishment and development of the infant gut microbiome. JAMA Pediatr. 2017, 171, 647–648. [CrossRef] [PubMed] 8. Pokusaeva, K.; Fitzgerald, G.F.; van Sinderen, D. Carbohydrate metabolism in Bifidobacteria. Genes Nutr. 2011, 6, 285–306. [CrossRef] [PubMed] 9. Parche, S.; Beleut, M.; Rezzonico, E.; Jacobs, D.; Arigoni, F.; Titgemeyer, F.; Jankovic, I. Lactose-over-glucose preference in Bifidobacterium longum NCC2705: glcP, encoding a glucose transporter, is subject to lactose repression. J. Bacteriol. 2006, 188, 1260–1265. [CrossRef] 10. Li, X.; Yin, J.; Zhu, Y.; Wang, X.; Hu, X.; Bao, W.; Huang, Y.; Chen, L.; Chen, S.; Yang, W.; et al. Effects of whole milk supplementation on gut microbiota and cardiometabolic biomarkers in subjects with and without lactose malabsorption. Nutrients 2018, 10, 1403. [CrossRef] 11. Szilagyi, A.; Shrier, I.; Heilpern, D.; Je, J.; Park, S.; Chong, G.; Lalonde, C.; Cote, L.-F.; Lee, B. Differential impact of lactose/lactase phenotype on colonic microflora. Can. J. Gastroenterol. 2010, 24, 373–379. [CrossRef] [PubMed] 12. Dallas, D.C.; Underwood, M.A.; Zivkovic, A.M.; German, J.B. Digestion of protein in premature and term infants. J. Nutr. Disord. Ther. 2012, 2, 112. [CrossRef] 13. Chaturvedi, P.; Warren, C.D.; Buescher, C.R.; Pickering, L.K.; Newburg, D.S. Survival of human milk oligosaccharides in the intestine of infants. Adv. Exp. Med. Biol. 2001, 501, 315–323. 14. Marcobal, A.; Sonnenburg, J.L. Human milk oligosaccharide consumption by intestinal microbiota. Clin. Microbiol. Infect. 2014, 18, 12–15. [CrossRef] [PubMed] 15. Sela, D.A.; Chapman, J.; Adeuya, A.; Kim, J.H.; Chen, F.; Whitehead, T.R.; Lapidus, A.; Rokhsar, D.S.; Lebrilla, C.B.; German, J.B.; et al. The genome sequence of Bifidobacterium longum subsp. infantis reveals adaptations for milk utilization within the infant microbiome. Proc. Natl. Acad. Sci. USA 2008, 105, 18964–18969. [CrossRef] 16. Borewicz, K.; Gu, F.; Saccenti, E.; Arts, I.C.W.; Penders, J.; Thijs, C.; Leeuwen, S.S.; Lindner, C.; Nauta, A.; Leusen, E.; et al. Correlating infant fecal microbiota composition and human milk oligosaccharide consumption by microbiota of 1-month-old breastfed infants. Mol. Nutr. Food Res. 2019, 9, 1801214–1801225. [CrossRef] [PubMed] 17. Ramani, S.; Stewart, C.J.; Laucirica, D.R.; Ajami, N.J.; Robertson, B.; Autran, C.A.; Shinge, D.; Rani, S.; Anandan, S.; Hu, L.; et al. Human milk oligosaccharides, milk microbiome and infant gut microbiome modulate neonatal rotavirus infection. Nat. Commun. 2018, 9, 1–12. [CrossRef] Microorganisms 2021, 9, 1153 16 of 17 18. Aakko, J.; Kumar, H.; Rautava, S.; Wise, A.; Autran, C.; Bode, L.; Isolauri, E.; Salminen, S. Human milk oligosaccharide categories define the microbiota composition in human colostrum. Benef. Microbes 2017, 8, 563–567. [CrossRef] [PubMed] 19. Moossavi, S.; Atakora, F.; Miliku, K.; Sepehri, S.; Robertson, B.; Duan, Q.L.; Becker, A.B.; Mandhane, P.J.; Turvey, S.E.; Moraes, T.J.; et al. Integrated analysis of human milk microbiota with oligosaccharides and fatty acids in the CHILD cohort. Front. Nutr. 2019, 6, 25. [CrossRef] [PubMed] 20. Gómez-Gallego, C.; Morales, J.M.; Monleon, D.; du Toit, E.; Kumar, H.; Linderborg, K.M.; Zhang, Y.; Yang, B.; Isolauri, E.; Salminen, S.; et al. Human breast milk NMR metabolomic profile across specific geographical locations and its association with the milk microbiota. Nutrients 2018, 10, 1355. [CrossRef] [PubMed] 21. Williams, J.E.; Price, W.J.; Shafii, B.; Yahvah, K.M.; Bode, L.; McGuire, M.A.; McGuire, M.K. Relationships among microbial communities, maternal cells, oligosaccharides, and macronutrients in human milk. J. Hum. Lact. 2017, 33, 540–551. [CrossRef] [PubMed] 22. Ruiz, L.; Espinosa-Martos, I.; García-Carral, C.; Manzano, S.; McGuire, M.K.; Meehan, C.L.; McGuire, M.A.; Williams, J.E.; Foster, J.; Sellen, D.W.; et al. What’s normal? Immune profiling of human milk from healthy women living in different geographical and socioeconomic settings. Front. Immunol. 2017, 8, 696. [CrossRef] 23. McGuire, M.K.; Meehan, C.L.; McGuire, M.A.; Williams, J.E.; Foster, J.; Sellen, D.W.; Kamau-Mbuthia, E.W.; Kamundia, E.W.; Mbugua, S.; Moore, S.E.; et al. What’s normal? Oligosaccharide concentrations and profiles in milk produced by healthy women vary geographically. Am. J. Clin. Nutr. 2017, 105, 1086–1100. [CrossRef] 24. Lackey, K.A.; Williams, J.E.; Meehan, C.L.; Zachek, J.A.; Benda, E.D.; Price, W.J.; Foster, J.A.; Sellen, D.W.; Kamau-Mbuthia, E.W.; Kamundia, E.W.; et al. What’s normal? Microbiomes in human milk and infant feces are related to each other but vary geographically: The INSPIRE Study. Front. Nutr. 2019, 6, 1–20. [CrossRef] 25. Lane, A.A.; McGuire, M.K.; McGuire, M.A.; Williams, J.E.; Lackey, K.A.; Hagen, E.H.; Kaul, A.; Gindola, D.; Gebeyehu, D.; Flores, K.E.; et al. Household composition and the infant fecal microbiome: The INSPIRE study. Am. J. Phys. Anthropol. 2019, 169, 526–539. [CrossRef] 26. Frank, J.A.; Reich, C.I.; Sharma, S.; Weisbaum, J.S.; Wilson, B.A.; Olsen, G.J. Critical evaluation of two primers commonly used for amplification of bacterial 16S rRNA genes. Appl. Environ. Microbiol. 2008, 74, 2461–2470. [CrossRef] [PubMed] 27. Carrothers, J.M.; York, M.A.; Brooker, S.L.; Lackey, K.A.; Williams, J.E.; Shafii, B.; Price, W.J.; Settles, M.L.; McGuire, M.A.; McGuire, M.K. Fecal microbial community structure is stable over time and related to variation in macronutrient and micronutrient intakes in lactating women. J. Nutr. 2015, 145, 2379–2388. [CrossRef] 28. Salter, S.J.; Cox, M.J.; Turek, E.M.; Calus, S.T.; Cookson, W.O.; Moffatt, M.F.; Turner, P.; Parkhill, J.; Loman, N.J.; Walker, A.W. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biol. 2014, 12, 87. [CrossRef] 29. de Goffau, M.C.; Lager, S.; Salter, S.J.; Wagner, J.; Kronbichler, A.; Charnock-Jones, D.S.; Peacock, S.J.; Smith, G.C.S.; Parkhill, J. Recognizing the reagent microbiome. Nat. Microbiol. 2018, 3, 851–853. [CrossRef] 30. Davis, N.M.; Proctor, D.M.; Holmes, S.P.; Relman, D.A.; Callahan, B.J. Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data. Microbiome 2018, 6, 226. [CrossRef] 31. Holmes, I.; Harris, K.; Quince, C. Dirichlet multinomial mixtures: Generative models for microbial metagenomics. PLoS ONE 2012, 7, e30126. [CrossRef] 32. Myatt, M.; Guevarra, E. Zscorer: Child Anthropometry z-Score Calculator; R Foundation: Vienna, Austria, 2019; p. 1. 33. Oksanen, J.; Blanchet, F.G.; Friendly, M.; Kindt, R.; Legendre, P.; McGlinn, D. Vegan: Community Ecology Package: R Package Version 2.4-6; R Foundation: Vienna, Austria, 2017. 34. McMurdie, P.J.; Holmes, S. Phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 2013, 8, e61217. [CrossRef] 35. Wickham, H. Ggplot2: Elegant Graphics for Data Analysis; Springer: Berlin, Germany, 2009. 36. Dufrene, M.; Legendre, P. Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecol. Monogr. 1997, 67, 345. [CrossRef] 37. Harrell, F.E. Hmisc: Harrell Miscellaneous: R Package; R Foundation: Vienna, Austria, 2020. 38. Wei, T.; Simko, V. R Package “Corrplot”: Visualization of a Correlation Matrix; R Foundation: Vienna, Austria, 2017. 39. Arumugam, M.; Raes, J.; Pelletier, E.; Le Paslier, D.; Yamada, T.; Mende, D.R.; Fernandes, G.R.; Tap, J.; Bruls, T.; Batto, J.-M.; et al. Enterotypes of the human gut microbiome. Nature 2011, 473, 174–180. [CrossRef] [PubMed] 40. Costea, P.I.; Hildebrand, F.; Arumugam, M.; Bäckhed, F.; Blaser, M.J.; Bushman, F.D.; de Vos, W.M.; Ehrlich, S.D.; Fraser, C.M.; Hattori, M.; et al. Enterotypes in the landscape of gut microbial community composition. Nat. Microbiol. 2018, 3, 8–16. [CrossRef] 41. Stewart, C.J.; Ajami, N.J.; O’Brien, J.L.; Hutchinson, D.S.; Smith, D.P.; Wong, M.C.; Ross, M.C.; Lloyd, R.E.; Doddapaneni, H.; Metcalf, G.A.; et al. Temporal development of the gut microbiome in early childhood from the TEDDY study. Nature 2018, 562, 583–588. [CrossRef] 42. Vatanen, T.; Plichta, D.R.; Somani, J.; Münch, P.C.; Arthur, T.D.; Hall, A.B.; Rudolf, S.; Oakeley, E.J.; Ke, X.; Young, R.A.; et al. Genomic variation and strain-specific functional adaptation in the human gut microbiome during early life. Nat. Microbiol. 2019, 4, 470–479. [CrossRef] [PubMed] 43. Ravel, J.; Gajer, P.; Abdo, Z.; Schneider, G.M.; Koenig, S.S.K.; McCulle, S.L.; Karlebach, S.; Gorle, R.; Russell, J.; Tacket, C.O.; et al. Vaginal microbiome of reproductive-age women. Proc. Natl. Acad. Sci. USA 2011, 108 (Suppl. 1), 4680–4687. [CrossRef] Microorganisms 2021, 9, 1153 17 of 17 44. Wu, X.; Jackson, R.T.; Khan, S.A.; Ahuja, J.; Pehrsson, P.R. Human milk nutrient composition in the United States: Current knowledge, challenges, and research needs. Curr. Dev. Nutr. 2018, 2, nzy025. [CrossRef] 45. Matheson, I.; Aursnes, I.; Horgen, M.; Aabø, Ø.; Melby, K. Bacteriological Findings and Clinical Symptoms in Relation to Clinical Outcome in Puerperal Mastitis. Acta Obstet. Gynecol. Scand. 1988, 67, 723–726. [CrossRef] 46. Cullinane, M.; Amir, L.H.; Donath, S.M.; Garland, S.M.; Tabrizi, S.N.; Payne, M.S.; Bennett, C.M. Determinants of mastitis in women in the CASTLE study: A cohort study. BMC Fam. Pract. 2015, 16, 1–8. [CrossRef] 47. Kvist, L.J.; Larsson, B.W.; Hall-Lord, M.L.; Steen, A.; Schalén, C. The role of bacteria in lactational mastitis and some considerations of the use of antibiotic treatment. Int. Breastfeed. J. 2008, 3, 6. [CrossRef] 48. Jost, T.; Lacroix, C.; Braegger, C.; Chassard, C. Assessment of bacterial diversity in breast milk using culture-dependent and culture-independent approaches. Br. J. Nutr. 2013, 110, 1253–1262. [CrossRef] 49. Delgado, S.; García, P.; Fernández, L.; Jiménez, E.; Rodríguez-Baños, M.; del Campo, R.; Rodríguez, J.M. Characterization of Staphylococcus aureus strains involved in human and bovine mastitis. FEMS Immunol. Med. Microbiol. 2011, 62, 225–235. [CrossRef] 50. Williams, J.E.; McGuire, M.K.; Meehan, C.L.; McGuire, M.A.; Brooker, S.L.; Kamau-Mbuthia, E.W.; Kamundia, E.W.; Mbugua, S.; Moore, S.E.; Prentice, A.M.; et al. Key genetic variants associated with variation of milk oligosaccharides from diverse human populations. Genomics 2021, 113, 1867–1875. [CrossRef] [PubMed] 51. Bode, L. Human milk oligosaccharides: Every baby needs a sugar mama. Glycobiology 2012, 22, 1147–1162. [CrossRef] 52. Viverge, D.; Grimmonprez, L.; Cassanas, G.; Bardet, L.; Bonnet, H.; Solère, M. Variations of lactose and oligosaccharides in milk from women of blood types secretor A or H, secretor Lewis, and secretor H/nonsecretor Lewis during the course of lactation. Ann. Nutr. Metab. 1985, 29, 1–11. [CrossRef] 53. Gay, M.C.L.; Koleva, P.T.; Slupsky, C.M.; du Toit, E.; Eggesbø, M.; Johnson, C.C.; Wegienka, G.; Shimojo, N.; Campbell, D.E.; Prescott, S.L.; et al. Worldwide Variation in Human Milk Metabolome: Indicators of Breast Physiology and Maternal Lifestyle? Nutrients 2018, 10, 1151. [CrossRef] [PubMed] 54. Velliyagounder, K.; Kaplan, J.B.; Furgang, D.; Legarda, D.; Diamond, G.; Parkin, R.E.; Fine, D.H. One of two human lactofer- rin variants exhibits increased antibacterial and transcriptional activation activities and is associated with localized juvenile periodontitis. Infect. Immun. 2003, 71, 6141–6147. [CrossRef] [PubMed] 55. Chowanadisai, W.; Kelleher, S.L.; Nemeth, J.F.; Yachetti, S.; Kuhlman, C.F.; Jackson, J.G.; Davis, A.M.; Lien, E.L.; Lönnerdal, B. Detection of a single nucleotide polymorphism in the human α-lactalbumin gene: Implications for human milk proteins. J. Nutr. Biochem. 2005, 16, 272–278. [CrossRef] 56. Lönnerdal, B. Effects of maternal dietary intake on human milk composition. J. Nutr. 1986, 116, 499–513. [CrossRef] [PubMed] 57. Bravi, F.; Wiens, F.; Decarli, A.; Dal Pont, A.; Agostoni, C.; Ferraroni, M. Impact of maternal nutrition on breast-milk composition: A systematic review. Am. J. Clin. Nutr. 2016, 104, 646–662. [CrossRef] [PubMed] 58. Institute of Medicine. Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids; The National Academies Press: Washington, DC, USA, 2005. 59. Pärnänen, K.; Karkman, A.; Hultman, J.; Lyra, C.; Bengtsson-Palme, J.; Larsson, D.G.J.; Rautava, S.; Isolauri, E.; Salminen, S.; Kumar, H.; et al. Maternal gut and breast milk microbiota affect infant gut antibiotic resistome and mobile genetic elements. Nat. Commun. 2018, 9, 1–11. [CrossRef] 60. Davenport, E.R.; Goodrich, J.K.; Bell, J.T.; Spector, T.D.; Ley, R.E.; Clark, A.G. ABO antigen and secretor statuses are not associated with gut microbiota composition in 1500 twins. BMC Genom. 2016, 17, 1–14. [CrossRef] 61. Seferovic, M.D.; Mohammad, M.; Pace, R.M.; Engevik, M.; Versalovic, J.; Bode, L.; Haymond, M.; Aagaard, K.M. Maternal diet alters human milk oligosaccharide composition with implications for the milk metagenome. Sci. Rep. 2020, 10, 1–18. [CrossRef] 62. Jantscher-Krenn, E.; Zherebtsov, M.; Nissan, C.; Goth, K.; Guner, Y.S.; Naidu, N.; Choudhury, B.; Grishin, A.V.; Ford, H.R.; Bode, L. The human milk oligosaccharide disialyllacto-N-tetraose prevents necrotising enterocolitis in neonatal rats. Gut 2012, 61, 1417–1425. [CrossRef] 63. Autran, C.A.; Kellman, B.P.; Kim, J.H.; Asztalos, E.; Blood, A.B.; Spence, E.C.H.; Patel, A.L.; Hou, J.; Lewis, N.E.; Bode, L. Human milk oligosaccharide composition predicts risk of necrotising enterocolitis in preterm infants. Gut 2018, 67, 1064–1070. [CrossRef] 64. Masi, A.C.; Embleton, N.D.; Lamb, C.A.; Young, G.; Granger, C.L.; Najera, J.; Smith, D.P.; Hoffman, K.L.; Petrosino, J.F.; Bode, L.; et al. Human milk oligosaccharide DSLNT and gut microbiome in preterm infants predicts necrotising enterocolitis. Gut 2020, 138. [CrossRef] 65. Taft, D.H.; Liu, J.; Maldonado-Gómez, M.X.; Akre, S.; Huda, M.N.; Ahmad, S.M.; Stephensen, C.B.; Mills, D.A. Bifidobacterial dominance of the gut in early life and acquisition of antimicrobial resistance. mSphere 2018, 3, 222–224. [CrossRef] 66. Hunt, K.M.; Preuss, J.; Nissan, C.; Davlin, C.A.; Williams, J.E.; Shafii, B.; Richardson, A.D.; McGuire, M.K.; Bode, L.; McGuire, M.A. Human milk oligosaccharides promote the growth of staphylococci. Appl. Environ. Microbiol. 2012, 78, 4763–4770. [CrossRef] 67. Goldsmith, S.J.; Dickson, J.S.; Barnhart, H.M.; Toledo, R.T.; Eiten-Miller, R.R. IgA, IgG, IgM and lactoferrin contents of human milk during early lactation and the effect of processing and storage. J. Food Prot. 1983, 46, 4–7. [CrossRef] 68. Koenig, Á.; de Diniz, E.M.A.; Barbosa, S.F.C.; Vaz, F.A.C. Immunologic factors in human milk: The effects of gestational age and pasteurization. J. Hum. Lact. 2016, 21, 439–443. [CrossRef] [PubMed] 69. Ninonuevo, M.R.; Park, Y.; Yin, H.; Zhang, J.; Ward, R.E.; Clowers, B.H.; German, J.B.; Freeman, S.L.; Killeen, K.; Grimm, R.; et al. A strategy for annotating the human milk glycome. J. Agric. Food Chem. 2006, 54, 7471–7480. [CrossRef] [PubMed]